Reproducibility of point-of-care ultrasonography for central vein diameter measurement: Separating image acquisition from interpretation.

PURPOSE: Central vein point-of-care ultrasonography must be reproducible to detect intravascular volume changes. We sought to determine which measurement step, image acquisition or interpretation, could be more compromising for reproducibility. METHODS: Three investigators each acquired inferior vena cava (IVC) and internal jugular (IJV) vein ultrasonographic sequences (US) from a convenience sample of 21 hospitalized general medicine participants and then interpreted each US three separate times. We partitioned the random errors of acquisition and interpretation, attributing wider dispersions of each to larger reductions in reproducibility. RESULTS: We analyzed 351 interpretations of 39 IVC and 432 interpretations of 48 IJV US. Reproducibility of the maximum (standard error of measurement 3.3 mm [95% confidence interval, CI 2.7-4.2 mm]) and minimum (4.8 mm [3.9-6.3 mm]) IVC diameter measurements were worse than that of the mediolateral (2.5 mm [2.0-3.2 mm]) and anteroposterior (2.5 mm [2.0-3.1 mm]) IJV diameters. The dispersions of random measurement errors were wider among acquisitions than interpretations. CONCLUSIONS: Among our investigators, central vein diameter measurements obtained by point-of-care ultrasonography are not sufficiently reproducible to distinguish clinically meaningful intravascular volume changes from measurement errors. Reproducibility could be most effectively improved by reducing the random measurement errors of acquisition. © 2017 Wiley Periodicals, Inc. J Clin Ultrasound 45:488-496, 2017.

Clinical measurements obtained from point-of-care ultrasound images to assess acquisition skills.

BACKGROUND: Current methods of assessing competence in acquiring point-of-care ultrasound images are inadequate. They rely upon cumbersome rating systems that do not depend on the actual outcome measured and lack evidence of validity. We describe a new method that uses a rigorous statistical model to assess performance of individual trainees based on the actual task, image acquisition. Measurements obtained from the images acquired (the actual desired outcome) are themselves used to validate effective training and competence acquiring ultrasound images. We enrolled a convenience sample of 21 spontaneously breathing adults from a general medicine ward. In random order, two trainees (A and B) and an instructor contemporaneously acquired point-of-care ultrasound images of the inferior vena cava and the right internal jugular vein from the same patients. Blinded diameter measurements from each ultrasound were analyzed quantitatively using a multilevel model. Consistent mean differences between each trainee's and the instructor's images were ascribed to systematic acquisition errors, indicative of poor measurement technique and a need for further training. Wider variances were attributed to sporadic errors, indicative of inconsistent application of measurement technique across patients. In addition, the instructor recorded qualitative observations of each trainee's performance during image acquisition. RESULTS: For all four diameters, the means and variances of measurements from trainee A's images differed significantly from the instructor's, whereas those from trainee B's images were comparable. Techniques directly observed by the instructor supported these model-derived findings. For example, mean anteroposterior diameters of the internal jugular vein obtained from trainee A's images were 3.8 mm (90% CI 2.3-5.4) smaller than from the instructor's; this model-derived finding matched the instructor's observation that trainee A compressed the vein during acquisition. Instructor summative assessments agreed with model-derived findings, providing internal validation of the descriptive and quantitative assessments of competence acquiring ultrasound images. CONCLUSIONS: Clinical measurements obtained from point-of-care ultrasound images acquired contemporaneously by trainees and an instructor can be used to quantitatively assess the image acquisition competence of specific trainees. This method may obviate resource-intensive qualitative rating systems that are based on ultrasound image quality and direct observation, while also helping instructors guide remediation.

Limited agreement between two noninvasive measurements of blood volume during fluid removal: ultrasound of inferior vena cava and finger-clip spectrophotometry of hemoglobin concentration.

OBJECTIVE: Plots of blood volume measurements over time (profiles) may identify euvolemia during fluid removal for acute heart failure. We assessed agreement between two noninvasive measurements of blood volume profiles during mechanical fluid removal, which exemplifies the interstitial fluid shifts that occur during diuretic-induced fluid removal. APPROACH: During hemodialysis we compared change in maximum diameter of the inferior vena cava by ultrasound ([Formula: see text]) to change in relative blood volume derived from capillary hemoglobin concentration from finger-clip spectrophotometry (RBVSpHb). We grouped profiles of these measurements into three distinct shapes using an unbiased, data-driven modeling technique. METHODS: Fifty patients who were not in acute heart failure underwent a mean of five paired measurements while an average of 1.3 liters of fluid was removed over 2 h during single hemodialysis sessions. [Formula: see text] changed -1.0 mm (95% CI -1.9 to -0.2 mm) and the RBVSpHb changed -1.1% (95% CI -2.7 to +0.5%), but these changes were not correlated (r -0.04, 95% CI -0.32 to +0.24). Nor was there agreement between categorization of profiles of change in the two measurements (kappa -0.1, 95% CI -0.3 to +0.1). SIGNIFICANCE: [Formula: see text] and RBVSpHb estimates of blood volume do not agree during mechanical fluid removal, likely because regional changes in blood flow and pressure modify IVC dimensions as well as changes total blood volume.

Biomarkers of inflammation and placental dysfunction are associated with subsequent preterm birth.

OBJECTIVE: To assess whether the analysis of high sensitivity C-Reactive Protein (hsCRP), a biomarker of inflammation, and placental growth factor (PlGF), a biomarker of placental dysfunction, could help identify patients at risk for preterm birth (PTB). METHODS: We performed a prospective cohort study of women with symptoms of preterm labor (22-33 6/7 weeks). Maternal serum was analyzed for hsCRP and PlGF. Median biomarker values were used as analytic cut-points. We performed chi-square tests of association between biomarkers and PTB, nonparametric tests to compare medians, and logistic regression to determine the odds of PTB associated with biomarker values. Test characteristics of each biomarker were calculated. RESULTS: 56.3% of the cohort (N = 96) delivered preterm. Median hsCRP (N = 78) was 4.34 mg/L, and median PlGF (N = 86) was 558.25 mg/l. In the setting of inflammation (high hsCRP), women with low PlGF had a 6.84-fold (95%CI: 1.57-29.80) increased risk of PTB. In the setting of placental dysfunction (low PlGF), women with high hsCRP had a 5.97-fold (95%CI: 1.52-23.43) increased risk of PTB. CONCLUSIONS: Our results suggest an interplay between inflammation and placental dysfunction in the pathogenesis of PTB. Analyzing biomarkers that reflect different pathways of PTB may hold promise for identifying patients at greatest risk.